Monday, February 27, 2012

Technology Ups and Downs - Part 2

Now we can see how our personal concerns mesh with levels of reliability as to whether we condemn a technology or not.  If we are only concerned about things happening to friends and family, and 'living history', then the odds of catastrophic failure need only be better than 1 in a thousand per year.  If we wouldn't use a technology because friends of a friend are killed, then the odds are around 1 in 10,000.  If we become terrified of something because of what we read in the newspapers, then the odds have to be better than 1 in a million.  We accept the very low odds of 1 in 10,000 for driving, but are very sensitive for other things.

We can take elevators next.  If I was the designer of the very first elevator, I would accept just a single cable driving the thing.  After all, I had personally never heard of a cable failing, and I'm too lazy to go up the stairs!  It's when we start mass deployment than these odds become untenable. Soon, we read about elevators crashing and killing people.  Nobody would step into one if they had the slightest concern, and the newspapers would have a feeding frenzy!  Now, we have a ridiculous number of cables holding the thing up, with multiple independent braking systems.  The odds have become better than 1 in a million.

The same went for the airliners of the 50's and 60's.  Who cared about crashing and dying at 1 in 1000, light up another cigarette!  Projections of mass deployment predicted a big crash every year.  Who would fly then?  They had to up their game, and put the odds to better than 1 in a million.  This was done with fanatically devotion to the quality of parts, and the frequent dismantling of engines.

So, in short, new technology was always hacked together at 1 in 500 to 1000.  Sensitivity rose with worldwide communication and mass deployment.  Still, the space shuttle flew at 1 in 500.

Now we come to nuclear plants.  When we were young designers, we knew all this.  We realized that standard engineering (hacking) could only give us 1 in 1000, so we designed independent backups, also at 1 in 1000, to give us a combined reliability of 1 in a million.  We were convinced we could never have a nuclear disaster with a commercial plant.

What went wrong?  How were we betrayed?  Experience now tells us we aren't doing better than 1 in 500 to 1000 per year, the same odds as personal (hacked) engineering!  The designers shiver in their graves! (or retirement).

-to be continued.

No comments: